New Modeling of Metal Oxide Surge Arresters

Transcription

1 Signal Processing and Renewable Energy September 2017, (pp.27-37) ISSN: New Modeling of Metal Oxide Surge Arresters Seyed Mohammad Hassan Hosseini 1 *, Younes Gharadaghi 1 1 Electrical Engineering Department, South Tehran Branch, Islamic Azad University, Tehran, Iran Accepted: 01 July, 2017 Abstract This paper describes simplified modeling of metal oxide surge arrester (MOSA) to operate analysis. This model is a new model proposed (P-K Model) to verify the accuracy in order to compare with IEEE and Pinceti Model. The simulations are performed with the Alternative Transients Program version of Electromagnetic Transient Program (ATP-EMTP). In the present paper, the MOSA models were verified for several medium voltages which consist of 18 kv and 21 kv, which 18 kv arrester was used in 22 kv system of Provincial Electricity Authority (PEA) and 21 kv arrester was used in 24 kv system of Metropolitan Electricity Authority (MEA) in Thailand. The P-K model was evaluate from different manufacturing, it is based on the General Electric (GE), Siemens and Ohio Brass as well as IEEE and Pinceti Model. The tests are performed by applying a fast front current surge with front time of up to 0.5μs and the standard impulse current surge (8/20μs). The results were compared between three models in order to calculate the error operation of the MOSA in the ATP-EMTP Program. The relative error of arrester models show that the P-K model can be used to simulate and calculate in ATP-EMTP program as well as IEEE and Pinceti model. In the case of fast front current surge, the P-K model has a maximum error of 5.39% (Ohio Brass, 10 ka, 21 kv) and has a minimum error of 0.24% (GE, 10 ka, 18 kv). Also, the standard impulse current surge, P-K model has a maximum error of 2% (Ohio Brass, 10 ka, 18 kv) and has a minimum error of 0.32% (Siemens, 10 ka, 21 kv) in the voltage response. Keywords: Oxide Surge Arrester, Frequency-Dependent Model, Lightning and Overvoltage. 1. INTRODUCTION The metal oxide varistor (MOV) material [1] used in modern high voltage surge arresters has a highly non-linear voltage versus current characteristic. The V-I characteristic is dependent upon wave shape of the arrester current. The physical construction of modern high voltage surge arresters consists of metal Oxide discs in- *Corresponding Author s smhh110@azad.ac.ir

2 28 Hosseini1. Gharadaghi. New Modeling of Metal Oxide Surge side a porcelain or polymer insulator. quately reproduces these effects without A higher voltage is achieved by adding requiring excessive computing time, uses a disks in series. Higher energy ratings are trial and error procedure. achieved by using larger diameter discs or The purpose of this paper is to present a parallel columns of discs. The highly nonlinear V-I characteristic obviates the need rester (MOSA), and was conduct to a com- simplified model for metal oxide surge ar- for series spark gaps. The electrical characteristics are determined solely by the prop- models have been proposed to simulation parison of several models [3, 4]. These erties of the metal oxide blocks. these dynamic characteristics. The results The ATP EMTP, Alternative Transients show that all models have similar performance when subjected to fast front surges Program version of Electromagnetic Transient Program [2], program allows the current and standard impulse current surge. modeling of this non-linear resistance through the ZnO Fitter routine and the 2. THE IEEE MODEL Type 92. Laboratory test data of metal oxide arrester discharge voltage and current The IEEE model was recommended by IEEE W.G [3], is shown in Fig.1. The A0 have indicated that the arrester has dynamic characteristics that are significant for and A1 are the two non-linear resistances and they are separated by a RL filter. For arrester studies involving fast front surges, which discharge currents with slow rising time, the are not well represented by the ATP model influence of the filter is negligible; thus A0 previously mentioned. Technical data show and A1 are essentially parallel and characterize the static behavior of the MOSA. For fast that for fast front surges, with rise time less than 8µs, the voltage waveform peak occurs front surge currents, the impedance of the before the current waveform peak and the filter becomes more significant, indeed the residual voltage across the arrester increases inductance L1 derives more current into the as the time to crest of the arrester discharge non-linear branch A0. Since A0 has a higher current decreases. voltage for a given current than A1, the model generates a higher voltage between its in- The increase could reach approximately 6% when the front time of the discharge is put terminals, which matches the dynamic reduced from 8µs to 1.3µs. characteristics of MOSAs. According to [3], this peak can reach up to 12%. It may be pointed out that the voltage across the arrester is not only a function of the magnitude of the discharge current, but it is also dependent on the rate of increase. This fact is particularly important in lightning studies. Several models, at different voltage levels, have been proposed to represent the frequency dependent characteristic of metal oxide surge arresters. The model proposed by the IEEE Working Fig. 1. IEEE Frequency-dependent model. Group, although having the purpose of finding a mathematical model that ade-

3 Signal Processing and Renewable Energy, September The proposed curves for A0 and A1 are 3. THE PINCETI MODEL shown in Fig.2 [5]. The per-unit values are The PINCETI model [4] presents derives referred to the peak value of the residual from the IEEE model, with some minor voltage measured during a discharge test differences. By comparing the models in with 10 ka standard impulse current surge Fig.1 and Fig.3, it can be noted that: (Vr,8/20). These curves are to be adjusting to The capacitance is eliminated, since its get a good fit with the published residual effects on model behavior is negligible, voltages for switching surge discharge currents. The inductance L0 represents the in- the two resistances in parallel with the inductances are replaced by one resistance R ductance associated with the magnetic fields (about 1 MŸ) between the input terminals, in the immediate vicinity of the arrester. with the only a scope to avoid numerical The resistor R0 is used to avoid numerical troubles. oscillations when running the model with a The operating principle is quite similar digital program. The capacitance C0 represents the external capacitance associated to to that of the IEEE frequency-dependent model. The parameter definition will be the height of the arrester. Starting from the shown that the proposed model can be physical dimensions of the arrester, some easily defined by adopting the two following rules: formulas are given in [1] to calculate L0, R0, C0 and R1. The definition of non-linear resistors The parameter L1 has the most influence characteristics (A0 and A1) is based on the on the result and a formula, starting from the curves shown in Fig. 2. These curves derives from the curves proposed by IEEE physical dimensions, is also suggests in [1]. However this constitutes only an initial value W.G , and are referred to the peak and L1 should be adjusted by a try- and error value of the residual voltage measured procedure to match the residual voltages for during a discharge test with a 10 ka lightning current impulse (Vr,8/20); - to define lightning discharge currents published in manufacture s catalogue. This model can the inductances, the following equations give satisfactory results for discharge currents within a range of times to crest for 0.5 can be used (values are in μh): μs to 45 μs. L 1 V r1 V r8 1. T2 20. V 4 N (1) Fig. 2. Non-linear characteristic for A 0 and A 1. V r8 20 V r8 20 V r8 20 L V r1 T2. V n (2) where: Vn= is the arrester rated voltage Vr1/T2 = residual voltage at 10 ka fast front current surge (l/t2 μs). The decrease time is not explicitly written because different manufacturers may use different values. This fact does not cause any trouble, since the peak

4 30 Hosseini1. Gharadaghi. New Modeling of Metal Oxide Surge 4. THE PROPOSED MODEL (P-K MODEL) Fig. 3. PINCETI model. The proposed model (P-K model) is shown in Fig. 4 and derives from P-K model that in [4]. It is intended for the simulation of the dynamic characteristics for discharge currents with front times starting from 0.5 to 8 μs. As in [4], between the non-linear resistances A0 and A1 only the inductance L1, which is defined the inductance, the following equations can be used (values are in μh): Fig. 4. P-K model value of the residual voltage appears on the rising front of the impulse. Vr8/20 = residual voltage at 10 ka current surge with a 8/20 μs shape. The proposed criteria do not take into consideration any physical characteristic of the arrester. Only electrical data are needed. The equations (1) and (2) are based on the fact that parameters L0 and L1 are related to the roles that these elements have in the model. In other words, since the function of the inductive elements is to characterize the model behavior with respect to fast surges, it seemed logical to define these elements by means of data related to arrester behavior during fast surges. L 9 Vr V r8 T2 20 V r8 20. V n (3) The resistance (R) has about 1 M to install between the input terminals. 5. SIMULATION RESULTS The simulations were performed with ATP EMTP program. The fast front current surge and the standard impulse current surge of each model for the 0.5µs and 8/20µs were presented at 18 kv and 21 kv in Table 2 and 3 respectively. In these tables, the relative error (Hr) in % defined by (4). Technical data of several arresters are reported in Table 1. H r V rsimv rman V rman. 100 (4) where: Table 1. Technical data of the considered arrester. Manuf. Rate 0.5 µsec 8/20 µs Maximum Discharge Voltage 10 ka Voltage GE (kv) IR-kVcrest 3 ka 5 ka 10 ka 20 ka Siemens Ohio Brass

5 Signal Processing and Renewable Energy, September Table 2. Calculation residual voltage and relative error (18 kv arrester). I Model Manuf. Index 0.5 µsec 8/20 µs Maximum Discharge Voltage 10 ka IR-kVcrest 3 ka 5 ka 10 ka 20 ka IEEE GE Vr H r Siemens Vr H r Ohio Brass Vr H r PINCE GE Vr H r Siemens Vr H r Ohio Brass Vr H r P-K GE Vr H r Siemens Vr H r Ohio Brass Vr H r Table 3. Calculation residual voltage and relative error (21 kv arrester). Model Manuf. Index 5 µsec 8/20 µs Maximum 10 ka Discharge Voltage 3 ka 5 ka 10 ka 20 ka R-kVcrest IEEE GE Siemens Ohio Brass GE PINCETI Siemens Vr H r Ohio Brass Vr H r GE Vr H r P-K Siemens Vr H r Ohio Brass Vr H r

6 32 Hosseini1. Gharadaghi. New Modeling of Metal Oxide Surge Fig. 5. Relative error on residual voltage, 18 kv (a) GE; (b) Siemens; (c) Ohio Brass. Fig. 6. The arrester product of GE 10 ka, 18 kv (a) The fast front current surge (0.5μs); (b) The

7 Signal Processing and Renewable Energy, September standard impulse current surge (8/20μs). Fig. 7. The arrester product of Siemens 10 ka, 18 kv (a) The fast front current surge (0.5μs); (b) The standard impulse current surge (8/20μs). Vrsim : is the simulated residual voltage; Vrman : is the manufacturer s residual voltage The relative error on residual voltage with each manufacturer which consists of GE, Siemens and Ohio Brass at 18 kv is shown in Fig. 5. The residual voltage results of the fast front current surge (0.5µs) and the standard impulse current surge (8/20µs) for current amplitude of 10 ka at 18 kv are presented in Figs. 6, 7 and 8. The relative error on residual voltage with each manufacturer which consists of GE, Siemens and Ohio Brass at 21 kv is shown in Fig. 9. The residual voltage results of the fast front current surge (0.5µs) and the standard impulse current surge (8/20µs) for current amplitude of 10 ka at 21 kv are presented in Figs. 10, 11 and CONCLUSION In this paper, the dynamic behavior of metal oxide surge arrester models is simulated with fast front time of up to 0.5μs and standard impulse current surge (8/20μs) which consist of IEEE, Pinceti and P-K model. The.

8 34 Hosseini1. Gharadaghi. New Modeling of Metal Oxide Surge Fig. 8. The arrester product of Ohio Brass 10 ka, 18 kv (a) The fast front current surge (0.5μs); (b) The standard impulse current surge (8/20μs). Fig. 9. Relative error on residual voltage, 21 kv (a) GE; (b) Siemens; (c) Ohio Brass.

9 Signal Processing and Renewable Energy, September Fig. 10. The arrester product of GE 10 ka, 21 kv (a) The fast front current surge (0.5μs); (b) The standard impulse current surge (8/20μs).

10 36 Hosseini1. Gharadaghi. New Modeling of Metal Oxide Surge Fig. 11. The arrester product of Siemens 10 ka, 21 kv (a) The fast front current surge (0.5μs); (b) The standard impulse current surge (8/20μs). Fig. 12. The arrester product of Ohio Brass 10 ka, 21 kv (a) The fast front current surge (0.5μs); (b) The standard impulse current surge (8/20μs). simulations of MOSA models were performed with the ATP- EMTP program. These three modeling results were compared with the data reported on the several manufacturer s catalogue, it was given to demonstrate the accuracy of models. The simulations of P-K model have been shown that it can use acceptably with a fast front current surge and standard impulse current surge at 18 kv and 21 kv in PEA and MEA respectively. In the case of fast front current surge, the IEEE model has a maximum error of 8.03% (10

11 Signal Processing and Renewable Energy, September ka, 18 kv), the Pinceti model has a maximum error of 1.27% (10 ka, 18 kv), and P ,Vol 21, Number 1, February K model has a maximum error of 5.39% (10 [6] Donoho, D., "Compressed ka, 21 kv). And also, the standard impulse sensing", IEEE Transaction on current surge, the IEEE model has a maximum error of 7.43% (3 ka, 18 kv), the Pinceti model has a maximum error of 9.32% (20 ka, 18 kv), and P-K model has a maximum error of 7.6% (3 ka, 18 kv) in the voltage response. However, the P-K model can be used to simulate and calculate in ATP- EMTP program as well as IEEE and Pinceti model. [7] Information Theory, Vol. 52, No. 4, pp Thein Thanh Lam, "Optimization in L1- Norm for sparse recovery", Faculty of Mathematics and Natural Sciences University of Oslo, REFERENCES [1] Richard G. Baraniuk, "Compressive Sensing", IEEE Signal Processing Magazine,VOL.24, Issue: 4, July [2] Jia Li, Zhaojun Wu, Hongqi Feng, Qiang Wang, "Greedy Orthogonal Matching Pursuit Algorithm for Sparse Signal Recovery in Compressive Sensing", IEEE Region 10 Conference, pp. 1-4, [3] Emmanuel J. Candès and Michael B. Wakin, "An Introduction To Compressive Sampling", IEEE Signal Processing Magazine, Volume: 25, Issue: 2, March [4] Ying Liu and Dimitris A. Pados, "Compressed-Sensed-Domain L1-PCA Video Surveillance", IEEE Transactions on Multimedia, VOL. 18, NO. 3, March [5] Rui Wang_, Jinglei Zhang, Suli Ren, and Qingjuan Li, "A Reducing Iteration Orthogonal Matching Pursuit Algorithm for Compressive Sensing", Tsinghua Science and Technology, pp.